HyperAI

Language Modelling On Salmon

المقاييس

Background (Domain) Consistency
Background (Random) Consistency
Background Alignment
Gender Consistency
Room Consistency
Sentiment Alignment
Sentiment Consistency
Speaker Consistency

النتائج

نتائج أداء النماذج المختلفة على هذا المعيار القياسي

جدول المقارنة
اسم النموذجBackground (Domain) ConsistencyBackground (Random) ConsistencyBackground AlignmentGender ConsistencyRoom ConsistencySentiment AlignmentSentiment ConsistencySpeaker Consistency
spirit-lm-interleaved-spoken-and-written55.064.059.585.054.552.073.581.0
last-language-model-aware-speech-tokenization55.560.554.570.561.051.564.063.0
textually-pretrained-speech-language-models55.560.556.569.559.053.061.569.0
textually-pretrained-speech-language-models54.061.556.568.059.051.559.069.5
text-free-prosody-aware-generative-spoken57.066.053.588.553.555.540.583.0
last-language-model-aware-speech-tokenization56.061.053.068.562.553.565.064.5
textually-pretrained-speech-language-models55.060.554.570.062.051.561.571.0
spirit-lm-interleaved-spoken-and-written53.555.551.567.054.548.054.569.5